The engine is the same. The difference is who can use it, how many databases it can reach, and two capabilities that exist for specific risks: pharma identifiers and AI clients reading your data. Price is a separate question, and Is DataOps TestGen free? answers it.
Feature by feature, as of TestGen 5.134.0
Sources are the October 2026 pricing page, the TestGen product page, and the release posts linked in each row.
| Capability | Open source | Enterprise |
|---|---|---|
| Users | 1 | Unlimited, $100 per user per month |
| Database connections | 1 | Unlimited, $100 per connection per month |
| Projects | 1 | Unlimited |
| Tables and data volume | Unlimited | Unlimited |
| Profiling, 55 column characteristics | Yes | Yes |
| Hygiene detection, 32 detectors | Yes | Yes |
| Generated tests, 34 types, plus 19 configured types | Yes | Yes |
| Commercial Pharma tests, 12 types (August 27, 2026) | No | Yes |
| Monitors: freshness, volume, schema, custom metrics | Yes | Yes |
| Quality scorecards and data catalog | Yes | Yes |
| Every supported database, plus files and DuckDB | Yes | Yes |
| Files and DuckDB (5.134.0) | Yes | Yes |
| Table Comparison migration tests (5.134.0) | Yes | Yes |
| MCP server, 96 tools | Yes | Yes |
| Hide source data from AI clients (5.134.0) | No | Yes |
| Single sign-on | No | Yes |
| Role-based access control | No | Yes |
| PII masking | No | Yes |
| Custom branding | No | Yes |
| Self-hosted, Docker or Kubernetes | Yes | Yes |
| Support | Community, on GitHub | DataKitchen support |
| Releases | Community | Priority |
| License | Apache 2.0 | DataKitchen subscription |
What both editions share
Everything that finds problems. The open-source edition profiles, generates tests, runs hygiene detectors and monitors, scores quality, and exposes all of it to an AI client over MCP. DataKitchen’s Periodic Table of Data Quality puts every check on one page and marks the only Enterprise tile, the pharma pack.
That design is deliberate. A free edition that holds back its best tests is a demo. You can’t judge a data quality tool on a demo, and you shouldn’t have to buy it to find out whether it catches anything in your data.
What only Enterprise adds
Room for a team
More users, more connections, more projects. Open source stops at one of each, and Enterprise removes the cap on all three.
Security and access
Single sign-on, role-based access, PII masking, and custom branding. These matter once a data steward, an analyst, and an engineer all look at the same results and shouldn’t all see the same columns.
Two risk-specific capabilities
The Commercial Pharma pack validates NPI, NDC, DEA, ICD, HCPCS, CMS Place of Service, and NCT numbers against their specifications, which a generic length or pattern check never does. And the hide-source-data option lets you connect Claude, Cursor, or Copilot to a regulated database without sending the values in it to the model.
Which edition fits you
Use open source if you are one engineer on one database, or if you are evaluating. Your tests and profiles carry over when you upgrade, according to the pricing page, so evaluating on open source costs you nothing later.
Move to Enterprise when a second person needs to review results, when you add a second database, when you work with commercial pharma identifiers, or when an AI client will touch data you can’t show it. Any one of those is enough.